The proposed project is part of a research agenda investigating the relationship between cognition and eye movements in on-going visual tasks such as reading. The chief goal of the project is to account for the variability in the durations of eye fixations made during reading. Three specific models that take different approaches in accounting for fixation duration variability are to be formulated mathematically and their predictions tested. The models are: (1) a Two-State Model (McConkie, 1989) developed to account for the shape of the hazard curve that underlies the frequency distribution of the durations of eye fixations made during reading, (2) Morrison's (1984) attention-based model, and (3) a Saccade Pacing Model based on suggestions by Levy-Schoen (1981). The research proposed is partially motivated by our observations that small effects in averages of eye movement measures are often due to much larger effects occurring infrequently. In order to avoid the misinterpretations of data that can arise from using simple averages, it is necessary to develop a model-based form of data analysis that is capable of providing a more complete description of experimentally-produced effects in eye movement data. A goal of the project is to meet this need by developing a model that, when fit to a set of data, will yield estimates for psychologically interpretable parameters of the model. These parameter estimates can then serve as theoretically cleaner dependent variables. Research techniques in the proposed studies involve recording readers' eye movements and, where needed, making changes in the display during selected saccades so that the stimulus is different during one eye fixation than it was during the last. Frequency distributions of the durations of selected eye fixations from the experiments are to be fit with the models, as a way of testing the psychological interpretations of the models' parameters. Other predictions derived from the models will also be tested, as a way of evaluating the viability of their proposed mechanisms for eye movement control. Finally, guided by the results of the proposed studies, a new model will be developed and tested. The development of this model will advance our understanding of the basis for eye movement control and its relation to cognition.